Substitutes or complements?

-Donal Curtin

Wednesday 7th November 2007

Lte's say you're in the media business and you're about to go to market with your product. You've got most of the design thinking done, and now you're sitting in the big meeting about distribution channels. And there's a big bun fight going on about whether to add an internet version.

The guys from finance say the website version will be a big mistake. Most importantly, it'll cannibalise sales of the hard copy. And you can't charge properly for stuff over the net; nobody's made a subscription model work profitably, and illicit copying is rife. Plus it'll eat its head off in IT support costs. It's exactly what you'd expect the bean counters to say.

The guys from marketing are all for it. They say the web version won't cannibalise sales: it'll add to them. It'll expand the market for both. Remember the fuss when radio came along and the music companies thought free songs over the airwaves would wipe them out? And what happened? A massive boost to sales as listeners liked what they heard and went out and bought it. Plus, they say, we've got the data (marketing always has data supporting marketing's latest wheeze) that shows the target consumers buy both physical and digital versions. Just like radio and CDs.

As you fret over which way to go, it's probably safe to say employing an economist to help with your choice is not top of mind. At first sight it's not even obvious there is any economics involved, let alone any that might be of practical commercial use.

Yet the heart of the issue between finance and marketing is indeed an economic one: it's the question of whether the hard copy and the soft copy are (in economists' terminology) substitutes or complements. If they're substitutes, finance is right: people will buy one or the other, and if there's a free or cheap version on the internet, that's all they're going to do. If they're complements, marketing is right: people will buy them together. They'll sample from their computer, and they'll buy in the store.
So which are they? Marketing say they've got the smoking gun - their data on purchasing patterns in the industry shows that people buy both. They're complements. End of story.

Trouble is, the story that marketing thinks emerges from the data may be a dangerous illusion. Suppose, for example, I'm a total music junkie. I buy old vinyl. I have the radio going all the time. They know me by name at Really Groovy. My hard disc is chocka with stuff I've, um, acquired over the internet. It's obvious that I'm in there in marketing's data as someone who buys music in all sorts of ways. But all the data is saying is, I like lots of music. It's not saying, although marketing thinks it is, that I'm indifferent to where I get it from. Give me your free website and I might never buy anything else from you. There's no telling on the evidence so far.

All of which leads to some classy research done by a guy called Matthew Gentzkow from the University of Chicago Graduate School of Business and published in the latest American Economic Review. He's figured out a way to interrogate the data to come up with the right answer as to whether the web copy and the hard copy work together or against each other. Unless you like reading sentences like 'I assume vij has a J-dimensional multivariate normal distribution with a free covariance matrix', you don't need to know the details, but he's solved the 'music junkie' issue, and some others that also had the capacity to mislead you, and he's come up with a clean way to tell whether your products are substitutes or complements.

And if you're thinking this is all laboratory stuff and not real world, Gentzkow's applied it to the real case of the Washington Post and its website, Post.com.

He found that at first blush, using marketing's sort of approach, it looked like the Post and its website were complements. Looked at more carefully, they emerged as substitutes. Raising the price of the Post by ten cents, for example, would see people defecting to the (free) website. Closing down the website would see people moving back to the Post (around 27,000 readers would take the paper again).

If the finance guys are now looking smug, they oughtn't to be. While the paper and the website did indeed compete against each other, the impact they had on each other wasn't very large, which meant that even with some modest degree of cannibalisation of each other, it could still make sense to offer them both. And he also found (contrary to finance's worries) that a free-to-air website could pay its way. Although it lost money when it started, by 2004 advertising revenue had probably made the website profitable. And in any event it tends to cost more to collect small subscription amounts that they're worth.

Bottom line: make sure the marketing data is answering the right question. If in doubt, get an economist to torture the data until it confesses.

Comments from our readers

No comments yet